Nutrition
Forget the macro wars. The dietary patterns with the best long-term outcome data — Mediterranean, MIND, DASH — agree on the basics: lots of plants and fish, minimal ultra-processed food. The disagreements are about details that are tiny compared to those basics.
Eat a Mediterranean-style pattern — vegetables, legumes, fish, nuts, olive oil, whole grains, fruit — and keep ultra-processed food, sugary drinks and processed meat rare. Within that, get enough protein in midlife and later to protect muscle and bone, and eat earlier in the day rather than late; fasting improves blood-sugar control but does not beat ordinary calorie restriction for weight loss, and no human trial has tested it on lifespan. Most of the evidence on this page comes from large observational studies, which show associations; where randomised trials exist, the sections below say so.
The pattern matters more than any single food
The Mediterranean pattern is the best-evidenced way to eat for a long life. A 2024 review of 28 studies in older adults found lower all-cause and cardiovascular mortality with closer adherence.[1] The landmark trial, PREDIMED, randomised 7,447 Spanish adults at high cardiovascular risk and found roughly 30% fewer heart attacks and strokes.[2] Its 2013 paper was retracted and republished after randomisation irregularities affecting about a fifth of participants; the result survived re-analysis, but Dietary patterns rates the trial itself as moderate evidence.
The MIND diet, a Mediterranean–DASH hybrid built for the brain, is weaker than its reputation. Its famous 53% lower Alzheimer's risk came from the smallest study, 923 older adults;[3] pooled cohorts put the link at about a third of that, and a randomised trial in 604 older adults found no advantage over a control diet with the same mild calorie cut.[4]
Polyphenols — plant compounds in berries, olive oil, dark chocolate, tea and herbs — track with roughly 7% lower all-cause mortality across 7 cohorts and nearly 180,000 adults.[5] A PREDIMED re-analysis found 37% lower mortality in the highest-polyphenol fifth, but that compared self-selected groups, not randomised arms, and the link held only in the low-fat control arm.[6]
"Plant-based" alone is not the lever; food quality is. In observational data, a healthful plant pattern (whole grains, legumes, fruit, vegetables, nuts) tracks with lower mortality and an unhealthful one (refined grains, fruit juices, sweets) with higher mortality.[7] Wide macronutrient ranges appear compatible with longevity when food quality is high.
The pattern's staple foods each have their own page. Fish and seafood: about two servings a week, mostly oily fish, not fried, with low-mercury species chosen. Legumes and soy: a daily serving modestly lowers LDL cholesterol and helps blood-sugar control, and the soy scares mostly fail on the evidence. Nuts and seeds: a small daily handful tracks with fewer cardiovascular and all-cause deaths, without the weight gain the calories predict. Whole grains: about three servings a day in place of refined grains, with oats and barley nudging LDL cholesterol down. Fruit and vegetables: most of the benefit arrives by five portions (about 400 g) a day, eaten whole rather than as juice. Milk, yogurt and cheese: optional, roughly neutral for mortality, with yogurt and fermented milks looking best.
Habitual fermented milk — yogurt and the cultured milks — tracks with about 6–7% lower all-cause, cardiovascular and cancer mortality in pooled cohorts; for yogurt alone the cardiovascular link was not statistically significant, aged cheese came out weaker (about 3%), and kefir has no pooled estimate.[8] See Fermented foods.
Fat: what replaces it decides the outcome
Fat quality outweighs fat quantity. Swapping saturated fat for polyunsaturated fat cuts cardiovascular disease by roughly 30%, while swapping it for refined carbohydrate achieves nothing.[9] In randomised trials, cutting saturated fat reduces cardiovascular events but not deaths, and the benefit tracks the size of the cholesterol drop.[10]
In practice: eliminate industrial trans fats, cook mainly with extra-virgin olive oil, and get omega-3s from fatty fish twice a week or about 1 g a day of EPA and DHA, the two long-chain omega-3 fats in fish oil. At that dose an after-the-fact analysis of the DO-HEALTH trial found modestly slower biological-age clocks, in a trial that missed all six of its main clinical outcomes.[11] The seed-oil panic doesn't hold up: at home-cooking intakes the omega-6 case against them is disconfirmed, and what harm exists belongs to repeatedly reheated commercial frying oil and the ultra-processed foods these oils arrive in. Dietary fats covers eggs and the omega-6 to omega-3 ratio.
Enough protein, especially past 65
Adequate protein in older adults preserves muscle, slows age-related muscle loss (sarcopenia) and improves recovery from illness or surgery. The international PROT-AGE consensus sets a floor of 1.0–1.2 g per kg of body weight a day for healthy older adults, 1.2–1.5 g/kg with acute or chronic illness, and up to 2.0 g/kg in severe illness, injury or malnutrition.[12] A floor is not a target: Protein argues for 1.0–1.5 g/kg past 65, and explains why the "low protein extends life" reading of animal data misfires in midlife and older adults.
Sugary drinks are the worst way to take sugar
Each daily 355 mL serving of a sugar-sweetened drink tracks with roughly 7% higher all-cause and 10% higher cardiovascular mortality in pooled cohorts.[13] Added sugar's link to an older biological-age clock reading, in 342 midlife women, was close to zero.[14] Sweeteners covers what added sugar, fructose and high-fructose corn syrup do.
Diet drinks are not a clean escape. In the Framingham cohort it was artificially sweetened soft drinks that carried a striking ischaemic-stroke signal, while sugared ones carried none; their dementia link did not survive adjustment.[15] The sugar alcohols erythritol and xylitol, common in "keto" and "sugar-free" products, carry the clearest cardiovascular warning sign among sweeteners: high blood levels track with more heart attacks and strokes, though no trial has tested intake. Keep them occasional, and avoid them as everyday sweeteners if you have heart disease, a past stroke, atrial fibrillation or a high clotting risk, or take antiplatelet drugs. Sugar substitutes covers the World Health Organization (WHO) and cancer-agency verdicts, and where stevia, monk fruit and allulose land.
Processed meat: a certain cause of cancer, a small absolute risk
The WHO's cancer agency (IARC) classes processed meat as a known human carcinogen, Group 1 — the same category as tobacco, which describes how sure the agency is that it causes cancer, not how much. Each 50 g a day, about two rashers of bacon, raises colorectal-cancer risk by about 18% in IARC's evaluation (16% in an independent pooling).[16] Against a lifetime baseline of about 3.9% for US adults, that moves an individual to roughly 4.6%, a rough calculation,[17] and the Global Burden of Disease project estimates that high-processed-meat diets account for about 34,000 cancer deaths a year worldwide, against roughly a million for tobacco.
Unprocessed red meat is a probable carcinogen (Group 2A), with about 15% higher colorectal-cancer risk in a 2025 pooling of 60 studies, published in a journal edited by its lead author.[18] It also tracks with roughly 11% higher cardiovascular risk per 100 g a day,[19] and with type 2 diabetes;[20] for dementia the signal is confined to processed meat.[21] Red and processed meat covers mechanisms, cooking methods and the 2019 NutriRECS panel that told people to carry on; Foods to limit gathers the full harm list.
Ultra-processed food: a large signal, softer than it sounds
A 2024 umbrella review of 45 meta-analyses across roughly 10 million people linked high ultra-processed intake to 32 adverse outcomes, but most were graded low or very low certainty and none high.[22] The defensible estimate is about 15% higher all-cause mortality in the highest intake group versus the lowest — the often-quoted 62% is a single-cohort outlier — and each extra 10% of calories tracks with roughly 3–10% higher mortality, depending on the analysis.[23][24] Healthy-user bias and reverse causation remain live explanations for the cohort data.
The strongest causal evidence is a 2025 randomised crossover trial in which both diets met UK dietary guidelines; the minimally processed one produced about twice the weight loss, a difference of about 1 kg.[25] It was 55 people over eight weeks, the ultra-processed diet was about a quarter more energy-dense, and three published critiques followed, so it points the way rather than settling it. See Ultra-processed food.
Carbohydrate quality: fiber over the glycemic index
Fiber is the best-evidenced marker of carbohydrate quality, graded moderate certainty against low-to-very-low for glycemic index. People eating the most have roughly 15–30% lower all-cause mortality, coronary heart disease, stroke, type 2 diabetes and colorectal cancer, with benefit continuing through 25–29 g a day and beyond.[26] The caveat is large: this is all observational, no randomised trial has shown that adding fiber prevents disease in ordinary adults, and one supplement trial increased the return of precancerous colon polyps. See Fiber.
Flattening after-meal blood sugar is a well-evidenced lever in diabetes; without diabetes the long-term benefit is unproven. People respond very differently to the same foods: an 800-person study predicted personal glucose responses from gut microbes, blood tests and habits,[27] though a 2026 re-analysis suggests much of that spread is day-to-day noise, and a 57-person Stanford study coined "glucotype" for a person's pattern of glucose swings.[28] Vegetables and protein before starch, intact whole grains and vinegar flatten the curve without cutting carbohydrate, while sourdough and eating earlier help less reliably; see Glycemic index.
Over-the-counter continuous glucose monitors read a lagged, brand-dependent signal, most of the "spikes" they flag are normal physiology, and no trial shows that reacting to them helps people without diabetes, while anxiety and over-restriction are real risks. A two-week experiment, not a lifelong wearable.
Timing: eat earlier; fasting doesn't beat eating less
Pooled trials rank an early eating window ahead of a late one for fasting insulin and body weight at matched calories,[29] but that ranking rests on indirect comparison, and the best head-to-head crossover since found no difference.[30]
Fasting and continuous calorie restriction produce similar weight loss over the long run. Pooling 99 randomised trials in 6,582 people, alternate-day fasting had a tiny edge of about 1.3 kg in short trials that vanished beyond 24 weeks;[31] a 2026 Cochrane review of 22 trials agreed.[32] Fasting covers alternate-day and fasting-mimicking diets.
Metabolic flexibility, the capacity to switch between burning glucose and fat, is a plausible aging-resilience marker, trained by the habits on this page rather than by anything specific to it.
GLP-1 drugs: real outcome benefit, with a muscle cost
GLP-1 receptor agonists — semaglutide (Ozempic, Wegovy), and the related tirzepatide (Mounjaro, Zepbound) — produce double-digit weight loss that lasts while you keep taking them. In adults with overweight or obesity and established cardiovascular disease, semaglutide cut heart attacks, strokes and cardiovascular deaths by about 20%: 6.5% had such an event over about three years, against 8.0% on placebo.[33] The benefit was unrelated to how much weight people lost, though about a third of it ran through the fall in waist size.[34] It also slowed several biological-age clocks, but in an after-the-fact analysis of 84 people with HIV-related fat redistribution, a fast-aging group, measured by the clocks' vendor.[35] Muscle and bone are lost alongside fat, much as with any large weight loss, so protein intake and resistance training matter. See GLP-1 receptor agonists.
Topics covered in depth
What to eat
- Dietary patterns — Mediterranean, MIND, DASH, Nordic and the Blue Zones compared.
- Protein — how much, from what, at what age.
- Dietary fats — the substitution effect, olive oil, eggs, seed oils.
- Fiber — a strong observational signal and a null trial record.
- Fermented foods — yogurt, kefir, kimchi, natto.
- A sample longevity week — seven days of meals that hit these targets, with variants.
What to limit
- Ultra-processed food — the classification, its critics, and the trials.
- Red and processed meat — the processed/unprocessed split and cooking method.
- Foods to limit — the full harm list, and what doesn't belong on it.
- Sweeteners — added sugar, fructose and high-fructose corn syrup.
- Sugar substitutes — artificial sweeteners, sugar alcohols, stevia, monk fruit, allulose and chicory syrup.
When and how you eat
- Fasting — time-restricted eating, alternate-day fasting, fasting-mimicking diets.
- Glycemic index — food order, vinegar, sourdough, and why the index misleads.
- Metabolic flexibility — switching between glucose and fat, and how it is trained.
- Continuous glucose monitors — what a sensor can tell a healthy adult.
Practical nutrition principles
- Default to a Mediterranean pattern: vegetables, legumes, fish, nuts, olive oil, whole grains, fruit; optional moderate dairy, modest poultry.
- Eat fish 2+ times a week (salmon, sardines, mackerel), or take EPA and DHA if not — see Omega-3.
- Hit protein targets: about 1.2–1.6 g/kg a day for active midlife adults and 1.0–1.5 g/kg past 65 (the upper half if you train), as about 0.4 g/kg per meal — roughly 25–30 g, or 30–40 g if you weigh more than about 75 kg — across 3–4 meals.
- Limit added sugar to about 25 g a day for women and 36 g for men (American Heart Association), stricter than the WHO's 10% of calories. The 2025–2030 US Dietary Guidelines set no daily ceiling but no more than 10 g in any single meal;[36] cap the daily total and avoid taking it in one hit.
- Limit ultra-processed food, starting with sugary drinks and processed meat.
- Keep unprocessed red meat to no more than about three portions a week (350–500 g cooked; one portion is about 120–170 g cooked) — less is fine, and the traditional Mediterranean pattern eats one or two — and keep processed meat occasional; replace it with legumes or nuts first, then eggs, poultry or fish, not refined carbohydrate.
- Eat 25–30 g of fiber a day from food. Getting from the typical 16 g to 25 matters far more than from 28 to 30.
- Keep sodium under about 2 g a day (5 g of salt), mostly by eating fewer packaged and restaurant meals, where most of it comes from. If your blood pressure is high, a potassium salt substitute at home is a reasonable swap, unless you have kidney disease or take a drug that holds on to potassium. See Salt and Salt substitutes.
- Front-load eating: don't skip or delay breakfast, make dinner the lighter meal, and finish eating around 19:00 and 2–3 hours before bed, in line with circadian biology. The bedtime gap is mainly about night-time reflux; a later supper hasn't been shown to worsen sleep itself (see Fasting).
- Don't smoke (smoking and nicotine); drink minimally (alcohol); hydrate adequately (water).
→ A sample longevity week puts all of these on a 7-day menu.
What's overhyped
- Carnivore, ketogenic and zero-carb diets for longevity — long-term outcome data are absent or unfavourable. In mice bred with a DNA-repair defect that speeds aging, near-doubling protein while cutting carbohydrate shortened median lifespan by 18% in males and 36% in females, with liver inflammation and injury signatures.[37] It is a fragile model and cuts both ways: halving protein extended lifespan in the males, which is not advice for humans either.
- "Detoxes" and cleanses — no clinical evidence; the liver and kidneys do this.
- Superfoods marketed for individual health claims — diversity beats specialisation.
- "Anti-inflammatory diet" supplement protocols — anti-inflammatory eating works through the pattern, not concentrated extracts.
For the full ordered action list drawn from every pillar, see the healthspan long list.